Main Loop hangs

[nikhil]

Guys,

I am facing a particular problem. I notice at times, the main loop stops functioning even though interrupts works fine. Any idea why this would happen. I've set traps for stacks, address, osi fail, etc, uart errors, nothing shows up. It has been particularly notice under following conditions:

a) While writing to UART 2 and an INT0 happens which actually makes a write to I2C.

Also, we notice at times the MCU resets on it own ..and nothing shows up in the traps or error bits.

Would appreciate inputs..

Thanks in advance.

[jpc]

as usual, show some code that demonstrates the issue. It sounds to me your main code somehwere gets in an endless loop.
Debugging ISR's can be real pain. without any code we can only speculate.

[nikhil]

jpc, thanks. The codes has run into many lines, I dont know which part to past.

But with your experience, would you think the following could be reason:

a) The place where the MCU Main loop stops, I've a delay_ms(10000) and I also have delay_ms(100) a couple of times in the interrupt, hence when the interrupt is called there could be a clash of the same delay_ms function ?

[jpc]

i would NEVER put any delay in interrupt, actually i try to avoid delays as proposed by the compiler completely, the main loop should run as fast as possible to get most out of your MCU. By using timers ( i usually have a 1 ms timer ISR running, this can provide many software timers/delay counters) instead of delays your MCU can do something usefull instead of wasting many cycles by a delay_ms( 10000) , imagine how much significant code could be executed by these.
Without knowing your interrupt-source, you also might have the next interrupt pending by the time the delay(s) allow returning from the ISR.

[nikhil]

Jpc, Thanks. As you notice I've only 4 months experience with this product against your 4 years..wow..your word is law.

from what I understood from your reply is..with a delay_ms(10000), the MCU actually stops for 10 seconds.. not allowing any other code to run and one should use timer delays when they want a delay in the interrupt routine and other routines.. and as far as possible avoid any delay in interrupts..

would you be able to share the code for timer delays..it would be highly appreciated..

[jpc]

what PIC are you using? The different family's have different interrupt vectors, for P30 as most simple example i start with something like this

Code: Select all

// using TMR3
//initialize  first by this code 
const  msecvalue       : word = clock_khz div 4;
var msecs : longint;
...
  msecs := 0;
  PR3 :=msecvalue;
  T3CON := %1000000000000000;
  IEC0.7 := 1;          // enable timer3 interrupts
...
procedure ms_tick_clock; org $22;
begin
  inc(msecs);
  IFS0.7 := 0;    // clear flag
end;

now you have a global msecs , inside the interrupt you may already make seconds,minutes/whatever you need .
In order to read these in your main, do make sure the interrupt will not interfere with the access if it is not atomic.

[nikhil]

thanks JPC, I use a PIC24FJ96GA010

[nikhil]

JPC, being honest.. I didnt really understand what you meant by "In order to read these in your main, do make sure the interrupt will not interfere with the access if it is not atomic."

[jpc]

first for your pic i think the vector is at $24 instead of the $22 on the P30.
In my example i use longint variable, if you read this fromthe main loop this will take more than one cycle so there is a risk that the ISR will occur between these introducing an error. The simplest solution is to disable interrupt while reading this variable, e.g.

Code: Select all

function get_time : dword;
begin
  IEC0.7  := 0;          // disable timer3 interrupts
  get_time := msecs;     // copy the value 
  IEC0.7  := 1;          // reenable timer3 interrupts
end;

[nikhil]

JPC it appears the program is moving out of the main loop, since if I remove the while true...wend loop the effect seems same. What I fail to understand is what event can make the code jump out of an endless loop

[jpc]

as you never showed any code it is impossible to solve this, in general the most unreliable element is the programmer him/herself
You are not using watchdog by accident ?

[nikhil]

JPC, I am using WDT and controlled via software and clearing flag within the main loop.

[jpc]

i would suggest not to use WDT until your application runs ok, as you have mentioned earlyer rather important delay's to be part of your code this might be your problem.

[nikhil]

Ok..lets try but the WDT is at 131 seconds and main loop takes less than 1 second to run expect when the 11 second delay is put to transfer data over gprs

[nikhil]

JPC, on second thoughts..I added the WDT code only to avoid this happening, hence I dont think this is the issue causing the jump out of main loop.